Conventional directed evolution involves discrete cycles of mutagenesis, transformation or in vitro expression, screening or selection, and gene harvesting and manipulation.[1,2] In contrast, evolution in nature occurs in a continuous, asynchronous format in which mutation, selection, and replication occur simultaneously. Although successful evolution is strongly dependent on the total number of rounds performed, [3] the labor- and time-intensive nature of discrete directed evolution cycles limit many laboratory evolution efforts to a modest number of rounds.
In contrast, continuous directed evolution has the potential to dramatically enhance the effectiveness of directed evolution efforts by enabling an enormous number of rounds of evolution to take place in a single experiment with minimal researcher time or effort. In a landmark experiment, Joyce and co-workers engineered a ribozyme self-replication cycle in vitro and used this cycle to continuously evolve a ribozyme with RNA ligase activity. This system remains the only reported example of continuous directed evolution and unfortunately cannot be easily adapted to evolve other biomolecules. [4-7]
Continuous directed evolution minimally requires (i) continuous mutagenesis of the gene(s) of interest, and (ii) continuous selective replication of genes encoding molecules with a desired activity. Several groups have developed methods to achieve continuous or rapid non-continuous cycles of mutagenesis.[8-13] For example, Church and coworkers recently developed multiplex automated genome engineering (MAGE), a system capable of generating targeted diversity in E. coli through automated cycles of transformation and recombination.[14] While these advances are capable of very efficiently creating gene libraries, they have not been linked to a rapid and general continuous selection, and consequently have not enabled continuous directed evolution.